Welcome to the third installment in my Analog Modeling blog series. In Part 1 I wrote about why equations are important for simulation. In Part 2 I suggested a process flow for turning device equations into a simulation model, and introduced the basic structure of a VHDL-AMS model. Now it’s time to begin the model definition process. As I outlined in Part 2, the first step is deciding what you want your model to tell you about the device, which may not be as obvious as it sounds.
The most important thing to remember when deciding what details you need to know about device behavior is this: calculations require CPU time, and CPU time directly affects how many seconds, minutes, and maybe hours will tick away on your wristwatch before the simulation is finished. Saying “I want to know everything there is to know about this device” is okay as long as you understand that the complexity of your model is directly proportional to the number of details you want to know about the device’s operation. I’ll say it again: more complexity means more of your time watching the simulation progress meter chug along. If you think about it, seldom do you really need to know every single detail about how a device operates. More than likely you will only be interested in 2 or 3 important performance metrics. Let’s use an incandescent lamp model as an example.
An incandescent lamp is designed to turn electricity into lumens. This transformation takes place by connecting a resistive wire, or filament (usually made of tungsten), to a source of electricity. As current passes through the filament, it heats up until it glows. The filament is enclosed in a glass bulb which is typically filled with an inert gas to reduce filament evaporation and oxidation. With this brief introduction in mind, here are some lamp properties that might be interesting to quantify in a simulation:
- Light intensity
- Electrical vs. thermal power
- Efficacy (Lumens per Watt)
- Filament temperature
- Rate of filament evaporation
- Rate of filament oxidation
If you create a lamp model to quantify all of these properties, your model will be quite complex, and potentially compute intensive. Following the analog modeling guidelines mentioned earlier, you need to narrow the options by deciding what you really want the model to tell you about the lamp’s performance. For our discussion, let’s focus on the relationship between the lamp’s electrical and thermal properties.
In my next blog post, Analog Modeling – Part 4, we’ll move to the next step in the analog modeling process and select equations that quantify the lamp’s interactions between electrical and thermal power.